Bad Modelling Job?
(February 24th, 2015) In recent years, the trillions of bacteria living in our guts have risen from obscurity to stardom. They are thought to affect almost everything, from our mood to health and disease. Studies often involve mice but are these rodents really an appropriate model?
The gut microbiome has been associated with an ever-growing list of diseases, including obesity, diabetes and even mental disorders such as anxiety and autism. Much like the Human Genome Project around 15 years ago, the booming microbiome research field has promised to deliver new revolutionary treatments, some as simple as eating a yogurt. Perhaps inevitably though, history repeats itself. After a few years of frantic microbiome sequencing and founding new biotech start-ups, microbiome researchers now have to face the hard questions: are the changes in the gut microbiome associated with certain diseases a cause or a consequence of the disease? How on earth can bacteria in the gut affect other parts of the body, such as the brain? What are the molecular mechanisms behind all this?
Studies with humans can at most reveal correlations between the microbiome composition and a given disease. For example: Bob is obese and happens to have a microbiome with lots of bacteria X, but John, who is slim, doesn’t. This suggests that bacteria X cause obesity, yet, there’s also a good chance that in fact it’s the other way round: obesity might somehow promote growth of bacteria X. Or maybe this type of bacteria thrives on Bob’s diet, or it simply prefers the unique environment of his gut.
It is virtually impossible, and unethical, to perform experiments in humans to explore causal hypotheses (does bacteria X cause obesity?) and control for confounding factors like diet and genetic background. Microbiome researchers have to use the next best thing: mice. There are, however, growing concerns within the scientific community that more often than not, data from mice can’t be extrapolated to humans for clinical purposes. Or at least, not easily.
In a new study Jeroen Raes and colleagues at the KU Leuven in Belgium carefully compared the human and mouse gut microbiomes to assess the strengths and pitfalls of this model system for studying microbiome-related diseases. “Microbiome research, notably its association to inflammatory diseases, relies heavily on mouse models […]. It is essential to know the qualities and limitations of each model to choose the correct one to test specific hypotheses,” says Sara Vieira-Silva, one of the authors conducting the study.
Mice are great for biomedical research. They share most of our genes and have similar anatomy and physiology. With the many available genetic tools, scientists can easily and quickly discover the function of literally any gene in the mouse genome, and recapitulate human disease in a controlled experimental set up. So where’s the catch? The problem is that although mice and humans share many similarities, there are also many differences.
Raes’ team performed comprehensive statistical analyses of all gut microbiomes from mice and humans published to date. These new data tell us what types of bacteria live in the gut in various scenarios (disease, diet, genetic background…), as well as their relative abundance. The team first compared the gut microbiomes of healthy humans and mice. And the differences start here.
Human and mouse guts have predominantly two ‘families’ of bacteria—Bacteroidetes and Firmicutes - but within these groups 85% of bacteria species found in mice are not present in humans. And the bacteria found in both? It appears their abundance in the gut also varies between mice and humans. The authors stress that many of these differences could simply be the result of technical limitations, like methodology or interference from external factors (diet, age, etc).
Currently, there are over 60 mouse models of Inflammatory Bowel Disease (IBD), but none fully recapitulates the disease. Even so, the changes in the gut microbiome of patients with IBD (when compared to healthy people) are similar to those observed in IBD mouse models. For example, there is a significant reduction in bacterial diversity in both IBD patients and IBD mouse models. However, some specific bacteria species will be more (or less) abundant in mouse but not in IBD patients. The same goes for obesity models. Overall, mice fed on high-fat diet, and also leptin-deficient mice, which cannot control their appetite, recapitulate the microbiome changes observed in obese people. But there are many discrepancies in the data, again likely due to external factors difficult to control, at least in human studies.
The conclusion? Well, mice are not people. Raes and colleagues warn microbiome researchers that extreme care should be taken when trying to extrapolate findings in mouse to humans. They should also make bigger efforts to standardise their protocols for animal handling and data analysis, and to share mouse models to eliminate any genetic variability that might skew the data.
“Most limitations of murine models for fundamental microbiome research can be overcome by methodical study design and statistical testing: either eliminating or keeping track of possible confounders (e.g. diet variation, genetic background) and testing for their influence on the results,” says Vieira-Silva.
Nevertheless, the authors conclude, when it comes to understanding the causes and molecular mechanisms behind human disease, mouse models seem to fit the bill. “Although the mouse microbiota composition is not identical to the humans’, most mechanisms of microbiota-host interaction will be shared between mice and humans,” concludes Vieira-Silva. “Mice models allow us to study these mechanisms with direct controlled experiments, towards the ultimate aim of providing therapeutic solutions.”